Pad effects on material-removal rate in chemical-mechanical planarization

Abstract

INTRODUCTIONAmong the newly developed planarization technologies for ultra-large-scale integration metallization, chemical-mechanical planarization (CMP) has emerged to be most promising because it can provide better local and global planarization of the wafer surface. 3 In recent years, CMP has emerged as an enabling technology for the next generation of chip manufacturing and has become the second fastest growing area of semiconductor-equipment manufacturing. Beside interlayer-dielectric planarization, CMP has also found applications in shallow-trench isolation, damascene technologies, 4,5 and other novel processing techniques, such as polishing of Si 3 N 4 balls for bearing applications. 6 In CMP, however, the manufacturing technique has outdistanced its underlying science. Current CMP practices focus more on empirical developments of polishing "recipes" for each specific application. This makes the CMP process design a trialand-error procedure. Moreover, valuable insights gained in one application cannot be readily used under a modified set of circumstances. Such lack of scalability and migratability of experimental data hinders the CMP process design efforts. The lack of physical understanding also makes the process difficult to control and results in suboptimal process performance. Accordingly, the present work aims at developing a mechanistic model of material removal during a CMP process by addressing the role of the employed porous pad.There exist several material-removal models for the CMP process. The oldest and most commonly used is Preston's model: 7 Ḣ ϭ C и P o и u, where Ḣ is the average thickness removal rate, P o is the applied pressure, u is the relative velocity between the pad and the wafer, and C is Preston's coefficient. Preston's equation is based on the observation in glass polishing and is an empirical model; however, it gives a good estimate of the material-removal rate (MRR) in CMP.Expressing the removal rate as a linear function of both normal and shear stresses, Wang et al. 8 and Tseng et al. 9 proposed a modification to the assumption of linearity inherent in Preston's equation. Using the analogy of the removal process to a traveling indenter problem and the principles of elasticity and fluid mechanics, they derived a feature-scale model with and u 1/2 dependence on the removal rate. Recently, Zhao and Shi 10 proposed a model of CMP based on elastic contact and soft-pad response. They proposed a dependence of for MRR and in-P 2/3 o P 5/6 o The role of a porous pad in controlling material-removal rate (MRR) during the chemical-mechanical planarization (CMP) process has been studied numerically. The numerical results are used to develop a phenomenological model that correlates the forces on each individual abrasive particle to the applied nominal pressure. The model provides a physical explanation for the experimentally observed domains of pressure-dependent MRR, where the pad deformation controls the load sharing between active-abrasive particles and direct pad-wafer contact. The pr...

Pad Surface Roughness and Slurry Particle Size Distribution Effects on Material Removal Rate in Chemical Mechanical Planarization

“…negative) particle diameters. The general behavior illustrated in Figure 1 follows from the fact that the total number of particles entrapped beneath a pad asperity is inversely proportional to the mean particle size (diameter) to the third power [4]. Thus, even though larger mean particles remove more material individually, there are far fewer of them.…”

“…Many models have been proposed to investigate the mechanisms of material removal in the CMP process. Features considered, include statistics of pad asperity height and spatial distribution [1], local deformation of individual cells [2], elastic contact between the wafer and the pad [3], and multi-level contact evolution at particle and asperity scales, leading to several domains of wafer/particle/pad contact [4]. Fu and Chandra [5] investigated the effects of viscoelastic pad properties on the material removal rate (MRR).…”

Performance and modeling of paired polishing process

“…The polishing pad is highly deformable porous solid with complex topology [24,25]. A modeling framework should embrace these unique geometric and kinematics features, even with an effective mean representation.…”

Pads for IC CMP

“…Fu et al [12] showed another nonlinear MRR model based on the concept of incomplete and complete contact between the wafer and the pad (beam bending model of pad). Bastawros et al [46] presented a model considering several saline features of the pad, such as the pad asperity of various amplitudes and frequencies, the local deformation of individual cells, the elastic asperity contact between the wafer and the pad, the multilevel contact evolution at the particle size scale, and the macro asperity scale. These factors lead to several domains of waferparticle-pad contacts.…”